• Geophysics and Geophysical Exploration
• /
• v.22 no.2
• /
• pp.80-87
• /
• 2019

In this paper, we propose a method to apply the polynomial fitting for regional-residual separation of microgravity data based on the characteristics of gravity anomaly without a prior information. Since the microgravity survey is usually carried out in small regions, it is common to approximate regional anomaly by the first-order polynomial plane. However, if the regional anomaly patterns are difficult to be approximated to a first-order plane, the complete gravity anomaly is divided into small zones enough to approximate first-order plane by means of Parasnis density estimation method. The regional-residual separation is then applied on the splitted zones individually. When the gravity anomalies can be splitted spatially, we showed that the residual anomalies can be more effectively extracted based on the regional geological structures by regional anomaly separation from each of the divided regions, rather than applying the entire data set at one time.

• Journal of radiological science and technology
• /
• v.46 no.4
• /
• pp.277-285
• /
• 2023

In this study, pneumonia identification networks with the small number of layers were constructed by using chest X-ray images. The networks had similar trainable-parameters, and the performance of the trained models was quantitatively evaluated with the modification of the network architectures. A total of 6 networks were constructed: convolutional neural network (CNN), VGGNet, GoogleNet, residual network with identity blocks, ResNet with bottleneck blocks and ResNet with identity and bottleneck blocks. Trainable parameters for the 6 networks were set in a range of 273,921-294,817 by adjusting the output channels of convolution layers. The network training was implemented with binary cross entropy (BCE) loss function, sigmoid activation function, adaptive moment estimation (Adam) optimizer and 100 epochs. The performance of the trained models was evaluated in terms of training time, accuracy, precision, recall, specificity and F1-score. The results showed that the trained models with the small number of layers precisely detect pneumonia from chest X-ray images. In particular, the overall quantitative performance of the trained models based on the ResNets was above 0.9, and the performance levels were similar or superior to those based on the CNN, VGGNet and GoogleNet. Also, the residual blocks affected the performance of the trained models based on the ResNets. Therefore, in this study, we demonstrated that the object detection networks with the small number of layers are suitable for detecting pneumonia using chest X-ray images. And, the trained models based on the ResNets can be optimized by applying appropriate residual-blocks.

• Proceedings of the KOSOMBE Conference
• /
• v.1992 no.05
• /
• pp.175-178
• /
• 1992

This paper proposes a method called linear prediction (a high performant technique in digital speech processing) for analyzing digital ECG signals. There are several significant properties indicating that ECG signals have an important feature in the residual error signal obtained after processing by Durbin's linear prediction algorithm. The ECG signal classification puts an emphasis on the residual error signal. For each ECG's QRS complex. the feature for recognition is obtained from a nonlinear transformation which transforms every residual error signal to set of three states pulse-cord train relative to the original ECG signal. The pulse-cord train has the advantage of easy implementation in digital hardware circuits to achive automated ECG diagnosis. The algorithm performs very well feature extraction in arrythmia detection. Using this method, our studies indicate that the PVC (premature ventricular contration) detection has a at least 90 percent sensityvity for arrythmia data.

• Geomechanics and Engineering
• /
• v.14 no.2
• /
• pp.115-129
• /
• 2018

This paper presents a set of results of plate load tests that imposed incremental cyclic loading to a sandy soil bed containing multiple layers of granulated rubber-soil mixture (RSM) at large model scale. Loading and unloading cycles were applied with amplitudes incrementally increasing from 140 to 700 kPa in five steps. A thickness of the RSM layer of approximately 0.4 times the footing diameter was found to deliver the minimum total and residual settlements, irrespective of the level of applied cyclic load. Both the total and residual settlements decrease with increase in the number of RSM layers, regardless of the level of applied cyclic load, but the rate of reduction in both settlements reduces with increase in the number of RSM layers. When the thickness of the RSM layer is smaller, or larger, settlements increase and, at large thicknesses may even exceed those of untreated soil. Layers of the RSM reduced the vertical stress transferred through the foundation depth by distributing the load over a wider area. With the inclusion of RSM layers, the coefficient of elastic uniform compression decreases by a factor of around 3-4. A softer response was obtained when more RSM layers were included beneath the footing damping capacity improves appreciably when the sand bed incorporates RSM layers. Numerical modeling using "FLAC-3D" confirms that multiple RSM layers will improve the performance of a foundation under heavy loading.

• Journal of Information Processing Systems
• /
• v.16 no.6
• /
• pp.1343-1358
• /
• 2020

In the stagewise arithmetic orthogonal matching pursuit algorithm, the weak threshold used in sparsity estimation is determined via maximum iterations. Different maximum iterations correspond to different thresholds and affect the performance of the algorithm. To solve this problem, we propose an improved variable weak threshold based on the stagewise arithmetic orthogonal matching pursuit algorithm. Our proposed algorithm uses the residual error value to control the weak threshold. When the residual value decreases, the threshold value continuously increases, so that the atoms contained in the atomic set are closer to the real sparsity value, making it possible to improve the reconstruction accuracy. In addition, we improved the generalized Jaccard coefficient in order to replace the inner product method that is used in the stagewise arithmetic orthogonal matching pursuit algorithm. Our proposed algorithm uses the covariance to replace the joint expectation for two variables based on the generalized Jaccard coefficient. The improved generalized Jaccard coefficient can be used to generate a more accurate calculation of the correlation between the measurement matrixes. In addition, the residual is more accurate, which can reduce the possibility of selecting the wrong atoms. We demonstrate using simulations that the proposed algorithm produces a better reconstruction result in the reconstruction of a one-dimensional signal and two-dimensional image signal.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.3E
• /
• pp.3-8
• /
• 1999

In speech recognition for real-world applications, the performance degradation due to the mismatch introduced between training and testing environments should be overcome. In this paper, to reduce this mismatch, we provide a hybrid method of spectral subtraction and residual noise masking. We also employ multiple model approach to obtain improved robustness over various noise environments. In this approach, multiple model sets are made according to several noise masking levels and then a model set appropriate for the estimated noise level is selected automatically in recognition phase. According to speaker independent isolated word recognition experiments in car noise environments, the proposed method using model sets with only two masking levels reduced average word error rate by 60% in comparison with spectral subtraction method.

• Proceedings of the KIEE Conference
• /
• 1999.07b
• /
• pp.627-629
• /
• 1999

This paper presents a new FDI scheme based on dynamic fuzzy model(DFM) for the nonlinear system. The dynamic behavior of a nonlinear system is represented by a set of local linear models. The parameters of the DFM are identified in on-line and aggregated to generate a residual vector by the approximate reasoning. The neural network classifer learns the relationship between the residual vector and fault type and used both for the detection and isolation of process faults We apply the proposed FDI scheme to the FDI system design for a two-tank system and show the usefulness of the proposed scheme.

• Journal of Korea Foundry Society
• /
• v.16 no.2
• /
• pp.141-148
• /
• 1996

In high chromium cast iron, the control of matrix microstructure as well as carbide structure is important to the performance as a wear resistant material. In this study, 3.0% C-24.0% Cr white cast irons with various molybdenum contents(residual, 1.0%, 3.0% and 5.0%) were solidified conventionally and unidirectionally for studying their effects on the microstructure and hardness. In the conventional casting, two sets of castings were poured from each melt. One set of the castings consisted of cylindrical bars of 10 and 20mm by 155mm long. The second set of the castings was a cylindrical bar of 30mm by 200mm long. On the other hand, a pep-set mold set on the Cu plate was employed to make the solidification unidirectionally. X-ray diffraction method was used to observe retained austenite and carbides in the high chromium cast iron. The morphology of eutectic $M_7C_3$ carbides changed from needle-like type to nodular type with the increase of Mo content. And, the presence of $M_2C$ carbides was identified in the sample where Mo was added over 3.0 %. Primary and eutectic carbides appeared as rod type and corngrain type, respectively in the unidirectionally solidified samples which were cut to parallel to the solidification direction. In the EDX analysis, Cr concentration was higher in the primary and eutectic $M_7C_3$ carbides, Mo in the $M_2C$ carbides, and Fe in the matrix.

• Transactions of Materials Processing
• /
• v.16 no.7
• /
• pp.509-515
• /
• 2007

As the springback of sheet metal during unloading may cause deviation from a desired shape, accurate prediction of springback is essential for the design of sheet stamping operations. When considering the case of a sheet metal being bent to radius $\rho$ that is such that the maximum stress induced exceed the elastic limit of the material, plastic strain in the outer surface will occur and the material will take a permanent set: but since, on removing the bending moment, the recovery of the material is not uniform across the thickness, springback will occur and the radius $\rho$ will not be maintained. Furthermore, when a tensile load being applied to each end of specimen, the tensile stress due to bending is increased and the compressive stress is decreased or cancelled and eventually the whole specimen may be in varying degree of tension. On the removal of the applied load the specimen loses its elastic strain by contracting around the contour of the block, the radius $\rho$ will be determined by the residual differential strain. Therefore in this study the springback is analytically estimated by the residual differential strains between upper and lower surfaces of greatest radius after elastic recovery, and a springback model based on the bending moment is also analytically derived for comparison purpose.

• Toxicological Research
• /
• v.34 no.2
• /
• pp.111-125
• /
• 2018

Solvents can be used in the manufacture of medicinal products provided their residual levels in the final product comply with the acceptable limits based on safety data. At worldwide level, these limits are set by the "Guideline Q3C (R6) on impurities: guideline for residual solvents" issued by the ICH. Diisopropyl ether (DIPE) is a widely used solvent but the possibility of using it in the pharmaceutical manufacture is uncertain because the ICH Q3C guideline includes it in the group of solvents for which "no adequate toxicological data on which to base a Permitted Daily Exposure (PDE) was found". We performed a risk assessment of DIPE based on available toxicological data, after carefully assessing their reliability using the Klimisch score approach. We found sufficiently reliable studies investigating subchronic, developmental, neurological toxicity and carcinogenicity in rats and genotoxicity in vitro. Recent studies also investigated a wide array of toxic effects of gasoline/DIPE mixtures as compared to gasoline alone, thus allowing identifying the effects of DIPE itself. These data allowed a comprehensive toxicological evaluation of DIPE. The main target organs of DIPE toxicity were liver and kidney. DIPE was not teratogen and had no genotoxic effects, either in vitro or in vivo. However, it appeared to increase the number of malignant tumors in rats. Therefore, DIPE could be considered as a non-genotoxic animal carcinogen and a PDE of 0.98 mg/day was calculated based on the lowest No Observed Effect Level (NOEL) value of $356mg/m^3$ (corresponding to 49 mg/kg/day) for maternal toxicity in developmental rat toxicity study. In a worst-case scenario, using an exceedingly high daily dose of 10 g/day, allowed DIPE concentration in pharmaceutical substances would be 98 ppm, which is in the range of concentration limits for ICH Q3C guideline class 2 solvents. This result might be considered for regulatory decisions.